Tuttle Tactical Management, LLC is an investment adviser registered with the U.S. Securities and Exchange Commission. You should not assume that any discussion or information contained in this letter serves as the receipt of, or as a substitute for, personalized investment advice from Tuttle ...
It is starting to look more and more like the recent decline was much ado about nothing. There are a couple of important events coming up over the next couple of days that could be catalysts one way or another:
1) Janet Yellen’s testimony on Thursday, and
2) The ECB meeting next Thursday March 6.
Over the intermediate term the trend in the market continues to move up, but in the short term it looks overbought and due for a selloff at some point. Treasury yields are sloping upwards and volatility has leveled out, both bullish signs. On the other hand, gold and oil are both in an uptrend. Overall we have returned to the market somewhat, and will continue to scale in if the uptrend continues.
Backtested Results: What to Watch Out For
Recently I have seen some money managers with some extremely impressive backtested performance results. In these cases, the results looked a little too impressive. I have long said, and will continue to say, that past returns are virtually meaningless, whether they are actual accounts, composites, or backtests. All they tell you is what happened in the past, they do not predict the future. What is most important about past returns is how they were achieved and whether they are repeatable. For example, I have seen a lot of impressive results for managers who use tactical strategies for high-yield bonds. These have been achieved during a massive bull market for bonds (which can’t continue for much longer). When bonds enter a bear market there will be no way these managers will be able to generate the same returns. That being said, I realize that investors scrutinize past returns when deciding how to invest. No matter what I say investors will continue to do that, so regarding backtesting, watch out for: results:
1) Backtesting with the benefit of hindsight. It is easy to generate great historical results through the benefit of hindsight. For example, after Apple had made its parabolic up move I created a momentum model to trade Apple stock. Of course the performance was off the charts, but it was useless because it was generated with the benefit of hindsight. I have also created commodity models that have extremely impressive results but if you look at where they came from, most are from the parabolic move in silver a few years back. A much better approach is to backtest with a forward-looking bias. For example, any bond methodology backtest that includes an inverse Treasury investment will not look as good as one that doesn’t. However, with interest rates more likely to rise than to fall, going forward with a bond approach that includes inverse Treasuries will probably do better than one that doesn’t.
2) Curve Fitting. This is finding the best set of parameters over a past period and expecting them to be the best set of parameters going forward, but this is rarely the case. You can generate a more robust set of parameters by walk-forward testing and using 3D analysis to look at your chosen parameters vs. nearby parameters.
3) Not Having a Premise. The market is ripe with all sorts of anomalies that you can easily find if you have enough computing power. There was the old Super Bowl anomaly: the every year that ends in a five is the market is up, etc. We once found an amazing anomaly in natural gas but we couldn’t figure out why it worked, so it was useless. The best approach is to start out with a premise that should work; for example it should make sense to buy stocks above their 200-day moving average and sell them below. You can then test the premise to see if it works “out of the box.” If it does, then you can use walk-forward testing and 3D analysis to further test it. If it still looks good you need to apply a forward-looking lens by asking why it worked, and is it likely to work going forward.